Water consumption has been rising during the last five decades across the globe because of population growth, higher standard of living, and economic development. Water resources limitation in arid and semi-arid regions, such as the most parts of Iran, has led to the emergence of sustainable water resource management. To deal with the complex and interlinked water challenges, we need knowledge about the quantity and quality of water in the watersheds through water balance investigation. Evaluating available water resources is a prerequisite step in the water resources management to avoid adverse environmental impacts. In this regard, water balance investigations, as a tool for quantitative evaluation of water resources, is a crucial issue in water resources management and water policy making process. Water balance can reveal the differences in income an outcome flow of the watershed and changes by the source of nature and human activities. However in the current study, we have faced some problems to calculate water balance such as: it is costly and time-consuming to provide reliable data for a large area and moreover the estimation methods of water balance components rely on the methods that would not be appropriate for Iran's meteorological, hydrological and hydrogeological conditions. For example, in many cases, the empirical methods and mathematical models used in water balance studies in watersheds of Iran were developed in other countries with different climates. Water balance models have been developed over multiple timescales (hourly, daily, monthly and yearly) and different spatial resolutions and varying degrees of complexity. Water balance models were fi rst developed in the 1940s by Thornthwaite. With the increasing use of water balance models to address hydrological problems, a considerable amount of efforts has been devoted to the development of such models and techniques. A variety of models have been considered, ranging from relatively complex conceptual models to complex physical based models. Previous studies showed that using conceptual water balance models rather than process-based models or data-driven models have some advantages. Process-based models require high resolution, in both space and time, data and data-driven models may not remain valid in changed climatic conditions. Therefore, we focused on distributed and semi-distributed mathematical conceptual models. After evaluation different models, we introduced and classify some models that have relative advantages for being applied in Iran. As the primary criteria to choose, these models have at least one of the following properties: they are conceptual models, developed for water balance (not the other purposes such as flood simulation), suitable for different regions; they have been used for arid and semi-arid areas; they use a simple method for water balance calculation; and their code, software or graphical interface are available freely. To choose and modify an appropriate water balance model for Iran's watersheds, we should consider the simpler one. This chose is regarding to the fact that as well as the lack of data, there are usually uncertainties about the model parameters and input data. The uncertainty could be due to sparse data and lack of knowledge about the natural systems. In the current study varied types of models from hydrological simulation and fully distributed physical models to the distributed and semi-distributed conceptual models were investigated. Among them, we considered the models that have a fewer number of parameters for calibration and consider hydrological spatial variability in basins. Due to the advantages and disadvantages of these models, ability of applying in the watersheds of Iran was also considered. Regarding to these facts, it seems the model developed by Portoghese et al. (2005) could be a good choice for the most arid and semi-arid regions in Iran, after necessary corrections and modifications. This model, regarding all hydrological processes, avoids unnecessary complexities. The model of Portoghese was successfully modified and coded in MATLAB software and implemented in different watersheds of Razavi Khorasan province, by the name of QDWB (Quasi Distributed Water Balance model). The developed model was named Qusi-Distributed according to fact that it does not use cell-to-cell flow routing. The model results, such as runoff, actual evapotranspiration, and deep percolation were in agreement with other studies, and consistent with the basin hydrological process. This modified model has been assessed in six plains of Razavi Khorasan province, during a research project.